engleski

Formant (F1-F3) and F0 gender differences between speakers of Croatian and Serbian RP

Most previous studies on Croatian and Serbian vowel system have been carried out on smaller groups of participants, deficient speech material or on heterogeneous groups of speakers (varying in their dialects, gender and age). Therefore, the aim of this study was to provide larger speech database for Croatian and Serbian, which could be useful for further forensic researches and also to test previous language differences between so closely related languages, using formant analysis. In 2015th Varošanec-Škarić and Bašić in their pilot study, acoustically described Croatian vowel system (based on F1-F3 values) for 14 male and 14 female speakers of Croatian RP. Marković and Bjelaković (2008) suggest that vowels [a], [e] and [u] are more peripheral in Serbian, than in Croatian, through contrastive analysis carried out on different corpora. Varošanec-Škarić, Bašić and Kišiček (2017) have made a contrastive comparison between male speakers of Croatian, Slovenian and Serbian (3 x 14 speakers) which showed significant differences between almost all vowels in the categories of open/closed, front/back and in the degree of lip protrusion and roundness. Therefore, it was necessary to analyze greater number of native speakers of Croatian and Serbian, of both genders (N=182) to test results from previous studies (with heterogeneous groups of participants) on greater corpora. Speech material for formant analysis presented in this study included 50 short sentences with 10 target words for each vowel ([i], [e], [a], [o], [u]). Formant and F0 measures have been taken from the stable part of the accentuated vowel. Speech data base also included 15 minutes of spontaneous speech (interview speech style) and 5 minutes of reading speech style, per each speaker (182) of both languages. Results Results of the previous study showed great number of language differences between Croatian and Serbian vowel system: vowels [i], [e], [o] and [u] proved to be significantly more closed in Croatian, vowels [e] and [i] significantly more front, and vowels [o] and [u] have been more rounded Bašić (2018). Formant values have been normalized using NORM normalizing packages (The Vowel Normalization and Plotting Suite ), with modified Watt and Fabricius method. Language differences have been also confirmed in F0 values. Croatian speakers showed significantly higher average F0 values compared to speakers of Serbian (male: 108 Hz and 118 Hz ; female: 179 Hz and 197 Hz). In this study we will represent analyzed gender differences that were examined through several acoustical parameters: formant values and their dispersion degree, F0 values and F0 range. Results confirmed the expected differences between male and female speakers in formant values (higher values for females in all vowels and formants, as well as in both languages), F0 values (40% higher values for female speakers of Croatian and 39% higher values for female speakers of Serbian, compared to male speakers of both languages). Statistical data verification indicated that female speakers have more dispersed vowel system (greater dispersion of formant values) in central and front vowels [a], [i] and [e], which was suggested by Gordon and Heath (1998). Frequency range has been calculated between speakers of same gender and different language as well as between different genders in same language (separately in Croatian and Serbian), according to the average absolute deviation calculations. The results showed that female speakers have significantly greater pitch range compared to male speakers in both analyzed languages.

Croatian, Serbian, Formants, Fundamental frequency, Gender difference

nije evidentirano